This post is part of a series about some experiments I’m doing combining space and time in slit scan photographs. The series starts here.
I’m beginning to get some images I like from the current slit scan series. I am experimenting with running the synthetic slit from right to left as well as the other was around.
I’m experimenting with varying the synthetic slit width (between 2 and 6 pixels seems to be best) and thus the synthethic scan time, so that the clouds are not completely weird, but don’t look natural, either.
During the programming to get a slit that I could move both ways, I made an error that produced this:
I thought the symmetry was interesting, and I made me wonder what things would look like If I used two synthetic slits, starting together in the center, and moving to the sides:
Except in the center, it’s missing the symmetry, because the clouds are different places at different times.
Oh, well.
Wonderful images!
Did you consider, instead of simply taking each column from a single image, using a weighted average? At this scale it’s impossible to see, of course, but at 100% it might make the clouds smoother. For example, for 5-pixel slits the weights might be:
0.8 0.2
0.6 0.4
0.4 0.6
0.2 0.8
0.0 1.0
I hope I’m clear…
Also, I’d be curious to see the results of horizontal slits and radial slits (e.g, take the center pixel from the first image, the pixels in a circle of radius 1 from the second image, the pixels in a circle of radius 2 from the third image, etc.)
Interesting idea about the weighting. Sort of like a slit with one (or two, if I extend the idea) fuzzy edge(s). I’d have to convert the images to floating point, and take out the gamma to do the processing right. After processing, I’d put the gamma back and convert to 16-bit per color plane precision. Done right, I could get some averaging of multiple frames for each column, and thus less noise in the shadows.
I have the ability to do a horizontal slit already, although I need to rework the code a bit to bring it up to the versatility that I now have with the vertical slit. I hadn’t thought of circular, tangential slit. I don’t think that would work when I’m just picking pixels from images, since there would be jitter in converting a perfect circle to the pixels available, but it could certainly work if I used weighting, as you suggested. I think that would be computationally cheaper than resing up and interpolating to get the pixel values.
I could do a radial sweep, too, but I think that will just look weird.
Lotta work ahead.
But that’s what’s fun.
Jim